Enhancement of discrete wavelet transform algorithm applied in medical image compression

ABSTRACT: The Discrete Wavelet Transform (DWT) is a widely used technique in medical image compression due to its ability to capture both spatial and frequency characteristics of images. However, traditional DWT suffers from several limitations, including the lack of phase information, shift variance, and limited directional selectivity, which can lead to distortions, misaligned edges, and loss of critical details in reconstructed medical images. This study proposes an enhanced DWT algorithm that addresses these limitations by integrating a trained Autoencoder for phase information preservation, implementing the Stationary Wavelet Transform (SWT) to mitigate shift variance, and employing a Directional Filter Bank (DFB) to improve directional selectivity. The proposed framework is evaluated using medical image datasets, including MRI, CT scans, and X-rays, with performance metrics such as Peak Signal-to-Noise Ratio (PSNR), Mean Reconstruction Error (MRE), and Coefficient Differences, Results demonstrate significant improvements in image quality, edge preservation, and compression efficiency, with up to 61.90% improvement in PSNR and reduced reconstruction error compared to traditional DWT. The enhanced algorithm ensures that critical diagnostic details are preserved, making it suitable for applications in medical imaging where accuracy and efficiency are paramount. This study contributes to the advancement of wavelet-based compression techniques, providing a robust solution for maintaining the integrity and diagnostic quality of medical images while reducing storage and bandwidth requirements.